The 18-Group Family Conglomerate Map

Eighteen family conglomerate clusters cover 33.4% of Thailand's SET Mainboard at the firm-count level — and 88.8% at the market-cap level. Every single SET50 firm sits inside one of those eighteen clusters.

Cluster membership is not a sample of the Thai market. It is the architecture of the Thai market.

This Note documents the cluster cohort — its size, its composition, its heterogeneity, and what cluster affiliation does and does not signal in the framework's reading. The headline finding, established up front and developed through the sections that follow, is that the eighteen clusters do not behave uniformly. Cluster B-axis means span twenty-four and a half points, from 58.0 at Minor International to 82.5 at the King Power cluster. Same family-control framing on paper. Different cluster-internal axis dynamics underneath.

That heterogeneity is the central reading. It is also the structural difference between Thailand's family-conglomerate canvas and Korea's chaebol-affiliate canvas, where cluster B-axis behavior is uniformly compressed. Note 7 of this series treats the Korea-Thailand B-axis comparison at full length. Note 4 establishes the cluster-level fact pattern that comparison rests on.

Universe coverage

The framework's curated family-alias dictionary maps 214 firms — 33.4% of the 641-firm SET Mainboard universe — to one of eighteen conglomerate clusters.^[1] The remaining 427 firms (66.6%) are non-clustered: family-, dispersed-, or foreign-parent-controlled issuers without alias-match attachment to any of the eighteen.

The market-cap distribution flips that ratio. Mean market capitalization in the clustered cohort is THB 78.8 billion; in the non-clustered cohort, THB 4.8 billion. The clustered cohort holds 88.8% of total SET Mainboard market capitalization — the eighteen clusters represent a third of firms but nearly nine-tenths of the index value.

The segment-level concentration is sharper still. Of the 28 Large-cap firms (≥ THB 100B), 27 sit inside one of the eighteen clusters. The single non-clustered Large-cap firm is Bank of Ayudhya (BAY), the MUFG-controlled Thai banking subsidiary that is foreign-parent-controlled rather than family- or state-cluster-affiliated.^[2] Of the 99 firms in SET100, 76 are clustered. Of the 49 firms in SET50, all 49 are clustered.

The 67% non-clustered cohort, by contrast, is overwhelmingly small-cap: 87.6% of non-clustered firms are below the THB 10 billion threshold. The non-clustered universe is the small-cap independent canvas that sits underneath the cluster-dominated headline.

Segment	Universe size	Clustered	Cluster share
SET50	49	49	100.0%
SET100	99	76	76.8%
Large-cap (≥ THB 100B)	28	27	96.4%
Universe market cap	THB 18.55T	THB 16.47T	88.8%

The 18-cluster map

The cluster table reads as the structural map of Thai listed-company architecture. Cluster size varies from 56 firms (PTT state-cohort) to 1 firm (Minor International). Axis profiles vary substantially — the heterogeneity reading developed in the next section.

Cluster	n	T	B	R	KS	Archetype mode	NDR ≥ 15%
PTT (state-cohort)	56	72.8	71.2	75.2	1	Chameleon (43%)	15 / 51
Bangkok Bank	36	69.2	71.9	73.4	1	Chameleon (42%)	2 / 21
Boonrawd	17	65.9	77.2	76.9	3*	Chameleon (41%)	0 / 14
KBank / Lamsam	14	70.1	70.3	72.9	0	Chameleon (50%)	0 / 8
Saha Pathanapibul	14	69.6	76.6	69.1	0	Hidden Gem (50%)	0 / 3
Central	12	73.0	71.2	75.6	0	Celestial (42%)	0 / 10
CP Group	10	72.3	62.5	71.7	1	Chameleon (60%)	4 / 6
King Power	8	64.5	82.5	71.1	1	mixed	0 / 6
Thai Summit	8	68.9	64.2	66.6	1	Chameleon (50%)	1 / 6
Royal-adjacent cohort	6	74.7	68.2	70.3	0	Chameleon (67%)	0 / 5
Mitr Phol	6	71.5	81.7	76.8	0	mixed	0 / 4
TCC / ThaiBev	6	72.0	72.2	70.2	1	Chameleon (50%)	0 / 3
Land & Houses	6	72.5	70.3	71.2	0	Chameleon (50%)	0 / 5
Osotspa	5	75.6	72.4	74.6	0	mixed	0 / 4
Thai Union	5	73.4	75.6	68.6	0	Chameleon (60%)	1 / 5
BDMS	2	77.0	79.5	82.5	0	Celestial	0 / 2
Sri Trang	2	75.0	66.5	69.5	0	mixed	0 / 2
Minor	1	72.0	58.0	67.0	0	Chameleon	0 / 1

*A methodological caveat on the Boonrawd cluster's 3 KS firms is necessary and is surfaced in the Boonrawd deep-dive below.

Cluster heterogeneity: the central reading

The eighteen clusters are not interchangeable. The B-axis range across clusters spans twenty-four and a half points — from 58.0 at the smallest cluster (Minor International) to 82.5 at the King Power cluster. The dispersion is not at the firm level; it is at the cluster level itself, where the cluster-aggregate B-axis mean varies almost as much as the universe-wide individual-firm B-axis dispersion.

The clusters distribute across three B-axis tiers.

High-B tier (B ≥ 75) — six clusters: King Power 82.5, Mitr Phol 81.7, BDMS 79.5, Boonrawd 77.2, Saha Pathanapibul 76.6, Thai Union 75.6. The high-B clusters share characteristics that reduce structural Balance-of-Power drag: only one of these six clusters has any firm at NDR ≥ 15% (Thai Union, with one such firm), and several of these clusters have transitioned to professionalized senior management while remaining family-controlled. Mitr Phol (3rd generation) and Boonrawd (4th generation) are both family-controlled with externally-recruited senior management at flagship operating units.

Mid-B tier (65 ≤ B < 75) — nine clusters, including Osotspa 72.4, TCC/ThaiBev 72.2, Bangkok Bank 71.9, Central 71.2, PTT 71.2, KBank 70.3, Land & Houses 70.3, the royal-adjacent institutional shareholding cohort 68.2, and Sri Trang 66.5. The mid-B tier is the governance-baseline tier: clusters that cluster around the universe-wide B-axis median of 74.0. Diverse ownership models sit here — state-enterprise hybrid (PTT), large family-controlled retail (Central), banking lineages (Bangkok Bank, KBank/Lamsam), and the institutional-shareholding cohort that operates with different lineage architecture from the family-conglomerate cohort.

Low-B tier (B < 65) — three clusters: Thai Summit 64.2, CP Group 62.5, Minor 58.0. The low-B clusters share the deepest NVDR exposure among the cohort. CP Group's cluster mean NDR of 17.16% is the highest of any cluster, and four of six CP firms with NDR data are above the 15% threshold. The low-B reading is structural: NVDR-mediated voting separation, intra-cluster cross-listing density, or both.

The structural conclusion follows from the table itself. Mitr Phol and CP Group are both Sino-Thai family-controlled multi-generational conglomerates with substantial listed footprints. Their cluster B-axis means differ by 19.2 points. Same family-control framing. Different cluster signatures.

CP Group: the low-B family case

Charoen Pokphand Group is the largest privately-controlled Thai conglomerate by global footprint and the framework's clearest example of a low-B family cluster. Ten firms in the alias-matched cohort; the strong-match subset (alias-group-aggregated holding ≥ 5%) includes CPALL, CPAXT, CPF, TRUE, GLAND, STANLY, CPT, and AYUD. T 72.3 / B 62.5 / R 71.7. The B-axis sits in the bottom three of the eighteen clusters.

The mechanism is a stack. Cluster mean NVDR Dilution Ratio is 17.16% — the highest of any cluster — and four of six firms with NDR data sit above the 15% threshold. CPALL (7-Eleven Thailand) and TRUE (Telecommunications) both register material NVDR exposure on the framework's scanner. Sector exposure spans Commerce (CPALL, CPAXT — Makro / Lotus's), Agribusiness (CPF — Charoen Pokphand Foods), Telecommunications (TRUE), Insurance (AYUD with Allianz JV), and Property Development (GLAND).

The cluster's archetype distribution is Chameleon-heavy: six of ten firms classified as Chameleon, two as Hidden Gem, one as Poison Apple, one as Kill Switch. The Kill Switch firm is GLAND (Grand Canal Land), flagged under the extended trading suspension cascade pathway.

Generationally, CP is in active transition. Senior chair Dhanin Chearavanont (born 1939) remains active at 86, while the third-generation operational transition has been underway across multiple operating units for several years. The Chearavanont family is Sino-Thai (Cantonese-origin), with the founding lineage tracing to Chia Ek Chor in early-twentieth-century immigration patterns to Thailand. The framework reads CP as the clearest "low-B family-cluster" case because both the NVDR exposure and the cross-listed structure across CPALL, CPAXT, CPF, and TRUE produce intra-cluster Balance-of-Power compression that other Sino-Thai clusters of comparable scale (Central, Boonrawd) do not exhibit.

PTT: the state-enterprise hybrid

The PTT cluster is the largest by firm count — 56 firms — and the only cluster with state-enterprise architecture. T 72.8 / B 71.2 / R 75.2. Mid-tier across all three axes. The Ministry of Finance holds 51.38% of the flagship PTT directly, with Vayupak Fund and group-affiliate holdings raising the aggregate state-cohort holding to 60.05%. Operating governance is overseen by the State Enterprise Policy Committee, with cabinet-level approval required for material transactions.^[3]

PTT's intra-cluster listed structure is the most extensive on SET. The flagship sits above PTTEP (exploration and production), GPSC (power generation), OR (oil retail), PTTGC and IRPC (chemicals), and TOP (refining), with several joint ventures and cross-listed affiliates extending the lattice further. Twenty-four of the cluster's 56 firms are classified Chameleon, fifteen Celestial, fifteen Hidden Gem, with one Poison Apple and one Kill Switch.

Cluster mean NDR is 12.15% with fifteen of fifty-one firms at NDR ≥ 15% — the second-deepest NVDR exposure among the eighteen clusters after CP Group. State-enterprise hybrids attract foreign-investor demand proportional to their large-cap index weight, and NVDR is the conduit through which that demand flows above the Foreign Ownership Limit ceiling on energy-and-utilities sectors.

The Kill Switch firm in the PTT cohort is THAI Airways International, currently undergoing court-supervised corporate rehabilitation under post-2020 bankruptcy proceedings. The KS-04 trigger reflects extended-period structural distress rather than a fast-moving event. THAI's governance restructuring through the rehabilitation process is itself a candidate for separate treatment in this series.

The cluster signature reading: PTT is the only state-enterprise cluster in the cohort, and it sits at mid-B because formal board independence is high (the flagship runs INED ratios above 78% on its own scorecard) while R-axis sits at 75.2 — the highest among the top-five clusters by firm count — reflecting a cleaner enforcement history and longer listing tenure than many family-controlled comparators.

Boonrawd: the alias-match caveat

The Boonrawd cluster carries seventeen alias-matched firms with cluster axis means T 65.9 / B 77.2 / R 76.9 — the lowest T-axis and one of the highest B-axes among the eighteen clusters. The cluster shows three Kill Switch firms in the production data, the highest count of any cluster. That headline number requires a methodological note.

The framework's cluster cohorts are defined by alias-token matching against curated dictionaries. Some matches are strong, indicating cluster-controlled holdings of at least 5% aggregated across the alias group. Others are weak: the alias dictionary picks up a token coincidence without supporting holding evidence. For cluster-level axis aggregation, the framework treats both strong and weak matches as cohort members.

Boonrawd's three Kill Switch firms — NRF (NR Instant Produce), PF (Property Perfect), and TR (Thai Rayon) — are all weak-token-only matches. NRF's alias-group-aggregated holding is 1.19%; PF's is 0.81%; TR's is 0.48%, and TR is in fact an Aditya Birla / Indorama-affiliated firm rather than a Bhirombhakdi-family-controlled entity.^[4] The strong-match Boonrawd subset — which corresponds to the actual Bhirombhakdi-family-controlled listed cohort, including S-Singha Estate, DTCENT, CSC, and SHR — has zero Kill Switch firms.

The cluster-level KS count of three overstates the Bhirombhakdi-family distress concentration. The strong-match reading is cleaner: Boonrawd is a fourth-generation Sino-Thai family-controlled brewery and diversified holding with a structurally healthy governance profile across its actually-controlled listed entities.

The methodological caveat applies more broadly. Several clusters carry weak-token-only matches in their cluster-aggregate tail. Cluster axis means are partially influenced by weak-match noise. Where the weak-match noise materially changes the structural reading — as it does for Boonrawd's KS count — the strong-match subset is the analytically appropriate cohort. The framework's published output exposes the alias-group-percentage attribution on every match for exactly this reason.

The flagship Boon Rawd Brewery, which holds the Singha brand, is itself unlisted; the framework's cluster cohort is composed of the family's listed real-estate, industrial, and diversified holdings.

Central Group: the balanced family case

The Chirathivat family's Central Group — twelve firms in the alias-matched cohort — is the framework's clearest example of a balanced family cluster. T 73.0 / B 71.2 / R 75.6. Uniformly mid-tier across all three axes with the most balanced profile of any of the top-five clusters. Cluster mean NDR is 4.16%, with no firms at NDR ≥ 15% — domestic-focused with controlled foreign-investor exposure architecture.

Central's archetype distribution is the most Celestial-heavy of the top-five clusters: five of twelve firms classified Celestial (42%), against a cluster-cohort average of 22%. Four Chameleons, two Hidden Gems, one Poison Apple. Zero Kill Switch firms.

Sector exposure spans Tourism and Leisure (Central Plaza Hotel, Centara), Property Development (Central Pattana — shopping-center operations and REITs), Commerce (Central Retail — department stores, supermarkets), and the agro affiliates of Central Restaurants Group. The third generation of the Chirathivat family holds active operational leadership; the fourth generation is in preparation. The family is unusually large for a Thai conglomerate cohort — public-record sources document fifty or more family members in operational roles across Central Group entities.

The Chirathivat lineage is Sino-Thai (Hainanese-origin), with founder Tiang Chirathivat (1900–1968) emigrating from Hainan in the 1920s. The cluster signature reads as the structural counter-example to CP Group: comparable Sino-Thai multi-generational family control, comparable listed-cohort scale, but the framework reads Central with stronger transparency outcomes (T 73.0 vs CP's 72.3), zero Kill Switch firms, zero high-NDR firms, and the highest Celestial archetype share among the top-five clusters. Domestic-focused architecture without the cross-border NVDR-pathway exposure that drags CP's Balance-of-Power axis.

TCC / ThaiBev: the founder-still-active case

The TCC Group cluster — Charoen Sirivadhanabhakdi's Thai-listed cohort — comprises six firms, with the strong-match subset spanning BJC (Berli Jucker), AWC (Asset World Corporation), and FPT (Frasers Property Thailand). T 72.0 / B 72.2 / R 70.2. Uniformly mid-tier; the cluster axis profile sits very close to the universe baseline.

Charoen Sirivadhanabhakdi (born 1944) is the founder and remains active chair at 80. Second-generation transition is imminent rather than completed; Thapana Sirivadhanabhakdi has been leading ThaiBev as CEO, but the parent-level governance structure remains in founder-active configuration. The Sirivadhanabhakdi lineage is Sino-Thai (Teochew-origin).

A structural feature distinguishes TCC from the other top clusters: the flagship beverage business, Thai Beverage (ThaiBev), is listed not on SET but on the Singapore Exchange. The Thai-listed TCC cohort consists of property and retail businesses (FPT and AWC for real estate, BJC for distribution) rather than the beverage flagship. This cross-border listing architecture — Thai-controlled business, Singapore-listed flagship — reflects a broader Sino-Thai pattern that the framework's Singapore coverage page treats separately. For Thailand-onshore reading, the SET-listed TCC cohort is the property-and-retail subset, not the full TCC business architecture.

Cluster mean NDR is 5.93%, with no firms at NDR ≥ 15% — moderate exposure profile. One Kill Switch firm sits in the cohort: NFC, a weak-token-only match (alias_group_pct of 2.20%) flagged under the trading-suspension-cascade trigger. The strong-match TCC subset has zero KS firms.

The cluster's structural reading is positional rather than dramatic: TCC sits at the universe-baseline on all three axes, with founder-active status signaling upcoming succession transition. Among the four imminent-succession founder-active clusters in the cohort — the others being Land & Houses (Anant Asavabhokin at 80), BDMS (Prasert Prasarttong-Osoth at 90), and CP Group (Dhanin Chearavanont at 86) — TCC is the one whose succession architecture is most actively visible in second-generation operational positioning.

The institutional-shareholding cohort

Six firms in the alias-matched cohort. T 74.7 / B 68.2 / R 70.3. Mid-tier all three axes, T highest of the three. Cluster mean NDR is 6.44%, with no firms at NDR ≥ 15%. Zero Kill Switch firms. Sector exposure aggregates across industrial conglomerate sectors and financial services.

The aggregate-only reading is the appropriate one for Note 4. The cohort's lineage architecture differs from the family-conglomerate cohort that the rest of this Note documents, and a deep-dive treatment is not within Note 4's scope. The cluster-row appears in the master table because it appeared in earlier framework outputs as a labeled alias group; its handling in this Note is limited to aggregate cohort statistics consistent with the framework's PUBLIC_GUARDRAILS protocol on aggregated institutional-shareholding cohorts.

Sino-Thai vs Thai-origin

Excluding the royal-adjacent institutional shareholding cohort and the PTT state-cohort — neither of which is family-founded — the cluster cohort distributes across sixteen family-cluster lineages. The cohort is dominantly Sino-Thai-immigrant-line. Ten of the sixteen family clusters trace founding lineages to Chinese-immigrant ancestry: CP Group (Cantonese), Bangkok Bank (Teochew), Boonrawd (Sino-Thai), KBank/Lamsam (Teochew), Saha Pathanapibul (Hokkien), Central (Hainanese), TCC (Teochew), Mitr Phol (Sino-Thai), King Power (Sino-Thai), and Osotspa (Sino-Thai). Five clusters are Thai-origin: Land & Houses, Thai Summit, Thai Union, BDMS, and Sri Trang. Minor International, the sixteenth family cluster, is Thai-naturalized American (William Heinecke), a non-Sino-Thai outlier within the family-cluster set.

The Sino-Thai family clusters represent 62.5% of the family-cluster cohort and approximately 70% of cluster-aggregate market capitalization. The pattern is consistent with the broader literature on Sino-Thai mercantile family dominance in Thai listed-company history — the immigration-and-business-formation patterns of the late nineteenth and early twentieth centuries that Wiwattanakantang (2001) and subsequent Thai family-firm research have documented.^[5]

The succession window

Four of the eighteen clusters carry founder-active or imminent-succession status, with founder ages above 80 in three of the four cases.^[6]

Cluster	Founder	Founder age (approx.)	Status
BDMS	Prasert Prasarttong-Osoth	90	Imminent
CP Group	Dhanin Chearavanont	86	Active transition
TCC / ThaiBev	Charoen Sirivadhanabhakdi	80	Imminent
Land & Houses	Anant Asavabhokin	80	Imminent

Note 4's role is to map the cohort-level demographic state, not to forecast specific transitions. The forward-watch list — which cluster's succession will trigger which type of governance event — is the subject of a separate Topical piece in this series. What the cohort-level reading shows is structural: four major clusters, all simultaneously at or beyond the typical generational-handoff window, distributed across property-and-retail (TCC), agribusiness-and-telecoms (CP), real-estate (Land & Houses), and healthcare (BDMS) — a non-trivial concentration of succession architecture in the 2025–2028 window.

Generation-axis correlation does not produce a clean directional reading. The B-axis range across founder-active clusters alone — from Minor's 58.0 to BDMS's 79.5 — spans more than twenty points, confirming that founder-active status is not the primary driver of cluster B-axis variation. The transition from founder to professional management at the second-and-third generation appears to compress B-axis variance toward the universe baseline, but with cluster sizes ranging from 1 to 56, that reading remains qualitative rather than quantitative.

Cluster vs non-cluster: what cluster membership signals

The framework's cluster-vs-non-cluster comparison produces a structurally important asymmetric finding.^[7]

Metric	Clustered (n=214)	Non-clustered (n=427)	Δ
T axis mean	70.91	67.04	+3.87
B axis mean	72.16	73.10	−0.94
R axis mean	73.35	73.36	−0.01
Composite g mean	72.26	71.38	+0.88
Celestial %	22.4%	15.0%	+7.4pp
Hidden Gem %	27.6%	37.9%	−10.3pp
Chameleon %	43.5%	36.3%	+7.2pp
Poison Apple %	2.3%	3.5%	−1.2pp
KS %	4.2%	7.3%	−3.1pp
Mean cap (THB B)	78.8	4.8	16.4×
Median cap (THB B)	8.1	1.9	4.3×

Cluster membership in Thailand correlates with higher T-axis (large-cap disclosure regimes, Big-4 audit ratios, SET100 / SETESG inclusion infrastructure), higher Celestial archetype share, and lower Kill Switch rate. It does not correlate with B-axis — clustered and non-clustered cohorts run within one point of each other on the Balance-of-Power axis. It does not correlate with R-axis — the gap is essentially zero.

Cluster membership is a transparency-and-distress signal. It is not a Balance-of-Power signal.

The asymmetry is structurally meaningful. If cluster membership were uniformly governance-protective on all three axes, the framework's reading of Thailand would mirror Korea's chaebol-affiliate pattern, where every chaebol affiliate sits near the same compressed B-axis number. Thailand's cluster signal is partial. It protects the disclosure-quality margin and reduces the Kill Switch hazard, but it does not move the Balance-of-Power axis — and the Balance-of-Power axis is where minority-shareholder protection lives.

The reading is honest. Family-cluster membership in Thailand does not solve the Balance-of-Power problem. It shifts the disclosure profile and reduces the distress hazard. Investors who treat cluster affiliation as governance-quality reassurance are pricing the wrong variable.

What the cluster map shows

Eighteen clusters. Twenty-four and a half points of B-axis variance across them. Eighty-eight percent of market capitalization. One hundred percent of SET50.

The structural takeaways are three.

First, family-cluster membership in Thailand is not interchangeable. Cluster identity matters more than cluster membership. The cluster B-axis range from 58.0 to 82.5 spans almost as much as the universe-wide individual-firm B-axis dispersion — and within that range, comparable Sino-Thai multi-generational family-controlled conglomerates of similar listed scale can sit more than nineteen points apart on the axis that matters most for minority-shareholder protection.

Second, Thailand's family-cluster canvas is dominantly Sino-Thai-immigrant-line — ten of sixteen family clusters trace founding lineages to Chinese immigration patterns of the late nineteenth and early twentieth centuries, with the Sino-Thai cohort representing approximately 70% of cluster-aggregate market capitalization. The structural feature is consistent with the broader Thai family-firm literature.

Third, cluster membership is a transparency-and-distress signal, not a Balance-of-Power signal. The framework's reading of cluster affiliation in Thailand is deliberately asymmetric: cluster firms score higher on T-axis and lower on Kill Switch hazard, but their B-axis behavior does not differentiate them from non-clustered comparators. Investors and ESG-rating systems that treat cluster affiliation as a uniform governance-quality marker are reading a partial signal as a complete one.

The cluster map is the architecture of the Thai listed market. The framework's reading of it is that the architecture matters — and that within the architecture, no two clusters look the same.

B-axis means across 18 family-conglomerate clusters. 24.5-point range from Minor International (58.0) to King Power (82.5).

---

The Apex G-Score framework currently covers 641 SET Mainboard listed companies as of the April 2026 production snapshot. Underlying data: FY2025 cross-section. Scoring under TBR v2.0 weights: Transparency 0.30, Balance of Power 0.30, Conflict-of-Interest Risk 0.40.